this article describes a system being developed to segment handwritten date fields on bank cheque images. the handwritten information extracted from the date zone is segmented into day, month and year, and a hypothesi...
详细信息
To improve weak classifiers bagging and boosting could be used. these techniques are based on combining classifiers. Usually, a simple majority vote or a weighted majority vote are used as combining rules in bagging a...
详细信息
A partial discharge (PD) signal in a transformer must propagate through the winding to its terminals before it arrives at a monitoring device, so the signal, on which the failure diagnosis is based, must undergo some ...
详细信息
ISBN:
(纸本)0780354605
A partial discharge (PD) signal in a transformer must propagate through the winding to its terminals before it arrives at a monitoring device, so the signal, on which the failure diagnosis is based, must undergo some distortion. In this work, the elemental impulse response of the winding is obtained by detecting the propagation of a pulse signal in the winding. By measuring the nanosecond scale rapid variation of the waveform of model discharges, the rule of the propagation of the discharge pulse signal in the winding under various conditions, including types and locations of the discharge and measuring points, is studied. Based on statistical comparison, the factors which have influence on the diagnosis of the discharge pattern are analyzed, which improves the exactness of the pattern diagnosis in practice.
the proceedings contain 92 papers. the special focus in this conference is on Pierre Devijver Lecture and Hybrid and Combined Methods. the topics include: Some notes on twenty one 21 nearest prototype classifiers;adap...
ISBN:
(纸本)3540679464
the proceedings contain 92 papers. the special focus in this conference is on Pierre Devijver Lecture and Hybrid and Combined Methods. the topics include: Some notes on twenty one 21 nearest prototype classifiers;adaptive graphical patternrecognition beyond connectionist-based approaches;current trends in grammatical inference;classifier's complexity control while training multilayer perceptrons;a framework for classifier fusion;image patternrecognition based on examples;a hybrid system for the recognition of hand-written characters;improving statistical measures of feature subsets by conventional and evolutionary approaches;selection of classifiers based on multiple classifier behaviour;the adaptive subspace map for image description and image database retrieval;adaptative automatic target recognition with SVM boosting for outlier detection;a multiresolution causal colour texture model;writer identification;syntactic patternrecognition by error correcting analysis on tree automata;offline recognition of syntax-constrained cursive handwritten text;structural classification for retrospective conversion of documents;segmentation of date field on bank cheques;grammars and discourse theory to describe and recognize mechanical assemblies;computation of the N best parse trees for weighted and stochastic context-free grammars;partitional vs hierarchical clustering using a minimum grammar complexity approach;encoding nondeterministic finite-state tree automata in sigmoid recursive neural networks;a structural matching algorithm using generalized deterministic annealing;alignment and correspondence using singular value decomposition;fast candidate elimination using machine learning techniques and efficient alignment and correspondence using edit distance.
In this paper, the sub-units in each stream are used and embedded in the multi-stream model. In this framework, sign language recognition system was implemented and evaluated. Experiments were carried out for 5177 Chi...
详细信息
ISBN:
(纸本)7801501144
In this paper, the sub-units in each stream are used and embedded in the multi-stream model. In this framework, sign language recognition system was implemented and evaluated. Experiments were carried out for 5177 Chinese signs. the real time isolated recognition rate is 95.1%. For continuous sign recognition, the word correct rate is 91.8%. this has shown that parallel multi-stream model is powerful for sign language recognition considering the problem of scalability.
A digital integrated system is developed to record partial discharge (PD) pulse and to recognize their patterns by waveforms. Special considerations on experiment circuit are given to electromagnetic shielding, system...
详细信息
ISBN:
(纸本)0780354605
A digital integrated system is developed to record partial discharge (PD) pulse and to recognize their patterns by waveforms. Special considerations on experiment circuit are given to electromagnetic shielding, system bandwidth and pattern variety. Both typical laboratory models and simulation models of stator winding are tested, which represent corresponding PD type in air, transformer oil artificial cavity of stator bar insulation and along surfaces of endwinding. the waveforms of discharge current of distinct models are recorded and their respective features are extracted by segmented time domain data compression method. Artificial neural network is applied to recognize different patterns. Several factors concerning recognition rate is discussed. the training and testing result shows the potentiality of applying waveform analysis in partial discharge recognition.
this paper applies patternrecognition method for localisation of the PD site in the oil-filled transformer by acoustic measurements. By dividing the transformer into N numbers of 20 X 20 X 20mm(3) sub-modules, the PD...
详细信息
ISBN:
(纸本)0780354605
this paper applies patternrecognition method for localisation of the PD site in the oil-filled transformer by acoustic measurements. By dividing the transformer into N numbers of 20 X 20 X 20mm(3) sub-modules, the PD site is localised by comparing the spatial distances of each sub-module to the acoustic sensors. the numerical cases indicate that the proposed method is general and practical while avoiding some drawbacks in current,algorithms. Moreover, the method improves the accuracy of localisation substantially when compared with conventional localization method.
We present an algorithm for extracting the surface skeleton of a 3D object from its D6 distance transform. the skeletal voxels are directly detected and marked on the distance transform within a small number of inspec...
详细信息
In this paper, a stochastic polynomial tone model is presented for tone modeling in continuous mandarin speech. In this model, the pitch contour is described by a stochastic trajectory. the mean trajectory is represen...
详细信息
ISBN:
(纸本)7801501144
In this paper, a stochastic polynomial tone model is presented for tone modeling in continuous mandarin speech. In this model, the pitch contour is described by a stochastic trajectory. the mean trajectory is represented by a polynomial function of normalized time while the variance is time varying. After that, an effective training and recognition algorithm is developed respectively. Also the problem of missing observation is discussed. Decision tree is employed to cluster the tone pattern variations, which are represented by proposed model. Many possible factors other than tone of neighboring syllables were taken into consideration when the decision tree was constructed. the experiments result shows that the tone recognition speed can increase more than 10 times while the recognition error rates decreased by 16% compared with traditional HMM tone model.
the use of plane graphs for the description of image structure and shape representation poses two problems : (1) how to obtain the set of vertices, the set of edges and the incidence relation of the graph, and (2) how...
详细信息
暂无评论